Contact Quality Score
Definition
Contact Quality Score
Contact quality score is the rating a reviewer gives a single customer interaction against a defined quality scorecard, usually expressed out of 100. It is the only contact center metric that judges how work was done rather than how fast.
Handle time, resolution, and abandonment all measure output. Quality score measures conduct: whether the agent verified identity, followed policy, showed the right tone, and left an accurate record.
The score is only as credible as the scorecard behind it. Two reviewers scoring the same call must land within a few points, or the metric measures the reviewer instead of the agent.
Key takeaways
- Contact quality score rates one interaction against weighted scorecard criteria, not an agent’s whole month.
- Calibration between reviewers is what makes the score comparable across a floor.
- Auto-fail items for compliance breaches sit outside the weighted average.
- Quality scores are coaching inputs first and performance-management inputs second.
How it works
Contact quality score is produced by scoring an interaction against weighted criteria, summing the weighted results, and expressing the total as a percentage. Most programmes review a sample of four to eight contacts per agent each month.
A typical scorecard splits into four groups, and the weightings tell you what the business actually values.
| Criterion group | Typical weight | What it checks |
|---|---|---|
| Compliance and verification | 30% | Identity checks, disclosures, consent |
| Resolution accuracy | 30% | Correct answer, correct system action |
| Communication and tone | 25% | Clarity, empathy, control of the call |
| Record keeping | 15% | Notes, disposition, follow-up actions |
Compliance items usually carry an auto-fail. A call that skips identity verification scores zero regardless of how well the rest went, because the risk is not proportional to the points.
Calibration sessions keep the metric honest. Reviewers score the same contact independently, compare results, and adjust their reading of each criterion until the spread narrows.
Sampling is the other weak point. Reviewing eight contacts from an agent who handled 900 gives a wide confidence interval, which is why speech and text analytics now pre-screen the pool rather than replacing the human review.
Weightings should be revisited every year. A scorecard built around a product that no longer exists quietly measures the wrong things, and agents learn to score well rather than to serve well.
Disputes need a route. Agents who can challenge a score — with the recording attached — keep the process credible, and the appeal rate itself says a lot about reviewer consistency.
Sample selection needs a stated rule. Random sampling gives a defensible average, targeted sampling finds risk faster, and most mature programmes run both against separate quotas.
Scores mean nothing without a coaching loop. The value sits in the conversation that follows the review, so programmes that publish scores without scheduling feedback produce resentment instead of improvement.
The discipline sits inside quality assurance and depends on call quality monitoring for its raw material — recordings, transcripts, and screen capture.
Formal excellence frameworks make the same argument at organisation level. The U.S. National Institute of Standards and Technology publishes the Baldrige Excellence Framework, whose 2026 revision is now available after nearly 40 years of use.
Customer-side measurement provides the reality check. The American Customer Satisfaction Index, which reports national satisfaction quarterly and published its Quarter 2, 2026 reading, exists precisely because internal scores can drift from customer opinion — see ACSI.
Examples
Scorecard design follows regulatory exposure and channel, so the same score means different things across sectors. Three cases show how banking, healthcare, and retail build and use the metric.
Regulated financial services weight compliance hardest. Mis-selling and disclosure failures carry auto-fails, and many banks review a higher sample on advice calls than on servicing calls.
Healthcare support centers weight accuracy and privacy. A single unverified disclosure of patient information is treated as a critical failure irrespective of the rest of the interaction.
Retail and travel centers weight tone and recovery. Because most contacts are complaints or changes, scorecards reward de-escalation and clear expectation-setting over strict script adherence.
Telecom providers score sales and service on separate cards. A retention call is judged on the offer made and the disclosure given, while a fault call is judged on diagnosis accuracy.
Outsourced providers run a two-layer review. The client quality team samples the provider’s own scored contacts, which turns calibration into a contractual activity rather than an internal one.
Related terms
Contact quality score is the output of a review process with its own roles, tooling, and downstream uses. The terms below cover who scores, what gets scored, and how the result is checked against customer opinion.
- Quality Assurance: the review function that designs scorecards and runs calibration.
- Quality Analyst: the role that scores interactions and feeds findings back to team leaders.
- Call Quality Monitoring: the recording and review activity that supplies the sample.
- Call Auditing: the deeper compliance-driven review of selected interactions.
- Customer Satisfaction Rating (CSAT): the customer’s own verdict, used to validate internal scores.
- First Contact Resolution: the outcome metric quality scores are expected to predict.
- Call Center Performance Evaluation: the wider appraisal process that quality scores feed into.
FAQ
What is a good contact quality score?
Most programmes set 85% as the pass mark and 90% or above as strong. The absolute number matters far less than whether reviewers are calibrated.
How many contacts should be reviewed per agent?
Four to eight per month is standard, with more for new starters and for agents on regulated queues.
Should agents see their own scores?
Yes, with the recording and the reviewer’s comments attached, because a score without evidence cannot be coached.
Can quality scoring be automated?
Analytics can screen and pre-score at scale, but human review still decides borderline and compliance-critical cases.
Does quality score predict customer satisfaction?
Only when the scorecard rewards outcomes rather than script compliance.
How should quality score feed into pay?
Use it as one input among several rather than the sole trigger, because a small monthly sample cannot carry that much weight alone.
Source partners building calibrated quality programmes can compare delivery models across Outsource Accelerator hubs.
Independent
